Beautiful preview

AI lead

Shape the intelligence that helps a rep✨ understand, decide and act for the represented.

This is a hands-on role building intelligence that helps people achieve their goals. You will use AI tooling to build capabilities, run experiments and ship features that help reps✨ reason, remember and act reliably. You will own the quality of that behaviour in real use, keeping the representative relationship independent of any one model and authority enforced outside it.

What you will do

  • Build the reasoning and tool-use capabilities that turn a person’s goals into useful action. Follow the work through from an experiment to dependable behaviour in production.
  • Design context and memory that preserve what matters to the represented, respect privacy and remain portable across models and implementations.
  • Build evaluations from real representative tasks. Measure goal completion, factual reliability, appropriate refusal and recovery when something goes wrong.
  • Investigate failed or stalled work with the Success manager. Turn recurring problems into capabilities that help the whole kindness deliver more goals.
  • Choose and combine models using evidence about quality, latency and cost. Keep the system able to adopt better models without trapping a person’s history or relationship in one provider.
  • Work with Platform and Trust to make tool use bounded and observable. A convincing model response must never substitute for verified authority.
  • Keep learning from new research and tools. Ship improvements when the evidence supports them and make their practical value clear to the team.

You might be right for this if

  • You have shipped AI systems that people rely on outside a demo and stayed responsible for their behaviour.
  • You are curious, driven and passionate about making AI useful. You can turn an uncertain idea into a clear experiment and a capability people use.
  • You understand model behaviour, retrieval, memory and tool use well enough to diagnose failures across the whole system.
  • You can build evaluations that expose weaknesses and resist the temptation to optimise for a flattering score.
  • You are hands-on with AI tooling and can explain the trade-offs behind what you ship. You value evidence over a convincing story.

You can be based anywhere in the world.

Explore our hiring DNA →

Apply

Share a link to something you have shipped, built or improved and tell us in about 200 words what you did, what changed and why this role matters to you. If your work is private, describe it without sharing confidential material.

Apply by email →

Updated: